Distributed Prediction Control with Server Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current control apparatuses face high processing loads when using large-scale algorithms for predicting control target values, leading to increased costs due to the need for high-performance CPUs and large storage devices, either in the apparatuses themselves or in external servers, which can result in inefficient resource allocation and high costs.
Innovation Solution
A control apparatus that collects prediction information, repeatedly predicts a first target value, and transmits it to a server for a second target value with higher accuracy, allowing for the management of prediction errors and setting of control target values, with shorter prediction intervals in the apparatus and longer transmission intervals to the server, optimizing processing loads and resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If large-scale algorithms such as deep learning are used to increase prediction accuracy, then prediction accuracy is improved, but processing load increases
Solution Approach 1:
The prediction system is segmented into two parts: a lightweight local prediction unit in the control apparatus that performs frequent predictions with low processing load, and a server-based prediction unit that performs accurate but computationally intensive predictions less frequently. This segmentation allows the system to achieve high prediction accuracy while maintaining low processing load during normal operation.
Solution Approach 2:
The system uses periodic action by having the control apparatus transmit prediction information to the server at predetermined intervals rather than continuously. This periodic transmission allows the server to perform accurate predictions at specific intervals while the control apparatus maintains continuous control with lighter processing requirements, thus balancing accuracy and processing load.
2Device complexity
If one server is provided for a large number of control apparatuses, then the number of servers is reduced, but processing load on the server increases
Solution Approach 1:
The prediction functionality is segmented between the control apparatus and the server. The control apparatus performs local predictions for immediate control needs, while the server provides periodic accurate predictions. This segmentation distributes the processing load, allowing one server to efficiently serve multiple control apparatuses without being overwhelmed by continuous prediction requests.
Solution Approach 2:
The control apparatus performs self-service by executing local prediction algorithms independently to generate control target values for normal operation. This self-service capability reduces the frequency and volume of requests to the server, thereby reducing the server's processing load while still allowing the server to provide periodic accuracy corrections.
3Manufacturing precision
If prediction processing is performed frequently, then control accuracy is improved, but processing load increases
Solution Approach 1:
The prediction processing is segmented into frequent lightweight local predictions at the control apparatus and less frequent accurate server-based predictions. This segmentation enables the system to maintain high control accuracy through frequent updates while keeping the processing load manageable by using simpler algorithms for frequent operations.
Solution Approach 2:
The system implements periodic action by having the control apparatus perform local predictions continuously for real-time control, while transmitting prediction information to the server at predetermined intervals for more accurate predictions. This periodic interaction between local and server-based predictions maintains high control accuracy without requiring the full processing power of server-level algorithms to run continuously.
Data Source
AI summary
A control apparatus includes a prediction unit configured to repeatedly predict a first target value based on prediction information; a transmission/reception unit configured to repeatedly transmit the prediction information to a server and receive a second target value having higher prediction accuracy than the first target value predicted by the server; a management unit configured to update a first error of prediction in the prediction unit based on the second target value and the first target value; and a setting unit configured to set a control target value based on the first target value and the first error. A first time interval in which the prediction unit repeatedly predicts the first target value is shorter than a second time interval in which the transmission/reception unit repeatedly transmits the prediction information to the server.